Tree-Based Classification of Tabla Strokes

This study attempts to validate the effectiveness of tree classifiers to classify tabla strokes especially the ones which overlap in nature. It uses decision tree, ID3 and random forest as classifiers. A custom made data set of 650 samples of 13 different tabla strokes were used for experimental purpose. Thirty-one different features with their mean and variances were extracted for classification. Three data sets consisting of 21,361, 18,802 and 19,543 instances respectively, were used for the purpose. Validation was done using measures like receiver operating characteristic curve and accuracy. All the classifiers showed excellent results with random forest outperforming the other two. The effectiveness of random forest in classifying strokes which overlap in nature is evaluated by comparing the known results with multi-layer perceptron.